Predicting Attacks with Deep Neural Networks
1 min readFeb 16, 2018
Earlier this week was the Anti-Nex release and today is the release of the first pre-trained models to defend software applications.
Here are the supported applications and model accuracies for predicting an attack:
- Flask: 89% accuracy
- React and Redux: 87% accuracy
- Vue: 83% accuracy
- Django: 70% accuracy
- Spring: 66% accuracy
The repository for building datasets and training deep neural networks using Keras and Tensorflow is on GitHub:
https://github.com/jay-johnson/antinex-datasets
Inside you can also find the first non-attack dataset for training deep neural network models to defend your own software applications, infrastructure and personal property from network exploits.
Call to Action
- I am exploring using sentiment analysis with a Keras embedding layer and would love to hear your thoughts on if this is a good choice for training models to predict an attack within tcp payloads.
- Help creating new attack and non-attack datasets. Let me know if you are interested.
Thanks for reading and I hope you find these models valuable.
Shields up and good luck out there!
Jay